As more and more nodes (e.g., routers) are added into a conventional communications network, the size of the network increases, and issues such as scalability and slow convergence may arise. In communication networks such as the Internet, an autonomous system (AS) may have a common routing policy (either in a single network or in a group of networks) that is controlled by a network administrator (or group of administrators on behalf of a single administrative entity, such as a university, a business enterprise, or a business division). Within the Internet, an AS comprises a collection of routers exchanging routing information via a common routing protocol. Each AS on the Internet may be assigned a globally unique number, which is sometimes called an AS number (ASN).
In a network comprising a single autonomous system (AS) with a single area, each node needs to be aware of the positional relationships (i.e., adjacencies) of all other nodes, such that all nodes may build a topological map (topology) of the AS. Nodes may learn about one another's adjacencies by distributing (i.e., flooding) link-state information throughout the network according to one or more interior gateway protocols (IGPs) including, but not limited to, open shortest path first (OSPF) or intermediate system (IS) to IS (IS-IS). Specifically, nodes engaging in IGPs may distribute their own link state advertisements (LSAs) describing their own adjacencies to all their neighboring nodes, which may forward the received LSAs to all their neighboring nodes (except the node from which the LSA was received). This may allow the LSA to be distributed throughout the network such that all network nodes become aware of one another's adjacencies, thereby allowing the various nodes to build topology graphs (e.g., link state databases (LSDBs)). LSAs may be distributed upon network initialization as well as whenever a network adjacency changes (e.g., a node is added/removed or a node/link fails). A network change may lead to every node in the network having to re-compute a shortest path to each destination, and to update its routing information base (RIB) and its forwarding information base (FIB). Consequently, as more nodes are added to a network, link state distributions and shortest path computations may begin to consume more and more network resources, such as bandwidth and/or processing time.
Through splitting a network into multiple areas, the network may be further extended. However, there are a number of issues when splitting a network into multiple areas. For example, dividing an AS into multiple ASs or an area into multiple areas may involve significant network architecture changes. For another example, it may be complex to setup a multi-protocol label switching (MPLS) traffic engineering (TE) label switching path (LSP) crossing multiple areas. In general, a TE path crossing multiple areas may be computed by using collaborating path computation elements (PCEs) through the PCE communication protocol (PCEP), which may not be easy to configure by operators since manual configuration of the sequence of domains is required. Further, the current PCE method may not guarantee that the path found would be optimal. For yet another example, some policies may need to be reconfigured on ABRs for reducing the number of link states such as summary link-state advertisements (LSAs) to be distributed to other routers in other areas. Thus, as an AS grows larger (e.g., comprising more and more nodes), scalability issues may arise, which may result from, for example, the inability of a large network to quickly compute a shortest path to every destination on each node, and efficiently manage or distribute network topology information. Consequently, larger networks may suffer from slower convergence. For example, larger networks may require a longer period to build or update topology graphs, during which time data may be misdirected or lost. Moreover, re-convergence in large networks may also be an issue, as the inability to timely recover from a fault in a node/link (or some other condition that changes a network adjacency subsequent to initialization) may disrupt network services, such as the traffic transported from node to node along the shortest path.